Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
1
1
« Cognitive Systems » Engineering at Hydro-Québec
Human-Computer Interaction, ECSE 424/542,Electrical and Computer Engineering, McGill University,
September 12th 2019 Jeremy Cooperstock
André Gascon
Technologies opérationnelles
\PreMcGill-19-Internet.ppt 2019-09-19 Hydro-Québec 2017
2
Introduction
Interface design is not just common sense.
One has to :• Learn not to rely on his own impressions• Know the human being• Know about interface technology• Use a proper design methodology
– based on task analysis and usability testing• Use the proper techniques
• And still not rely on his own impressions
Hydro-Québec 2017 (André Gascon)
2
3
Introduction
Don’t rely on your own impressions- Task-task-task (goals, sub-goals, information used); nature of expertise; always validate
Peripheral perception- Layout, navigation, schematic-diagram-spatial, beware of movement and blinking, . . .
Anticipation- Consistency, affordance, transparency; mental model and sufficient information for correct anticipation
Cognitive System : Human + Machine
Hydro-Québec 2017 (André Gascon)
4
Objectives
• That you understand the necessary attitude :
>>> Don’t rely on your own impressions. <<<
< It isn’t natural, it has to be learned hammered in > < Tullis . . . >
Discuss the advance exercise video Suggest what you should rely on.
• That you get a flavour of «Cognitive Systems» Engineering ( CSE) in industry :
Show an example of CSE applied to an industrial problem Describe what it’s like in practice, how it fits within IT
Hydro-Québec 2017 (André Gascon)
3
5
Outline• Design example : Context Users Observations Conjecture Design
– Questions– Advance exercise : Discussion Recommandations
• Cognitive Systems Engineering
• Major lessons learned : Methodology and techniques Design
• Articulation with the rest of IT
• Questions (at least 5 minutes at the end)
• Examples of CSE projects (if time permits)
• Support information : Definitions Suggestions Problems Future of CSE References
Questions during the lecture are OK (in the past, it helped me improve the lecture).If needed, I’ll be glad to answer questions after the lecture.
6
Design example
Alarm annunciator
From diagnosis to situation awareness
(circa 1995, but the lessons remain valid)
Design example
Hydro-Québec 2017 (André Gascon)
4
Hydro-Québec = Electric Power Utility
Production Consommation
(*)
Production Consommation
Pas de batterie, seulement un peu d’inertie : il faut en tout temps équilibrer la production avec la consommation.
< Si on allume une lumière, il faut qu’un peu plus d’eau coule dans la turbine >
Transmission
8
Electrical Power Network : Generation
62 hydroelectric power plants
Hydroelectric power plant
Hydro-Québec 2017 (André Gascon)
Design example
Context
< www.hydroquebec.com >
5
9
Electrical Power Network : Transmission
32,800 km of power lines550 substations
High-voltage substationHigh-voltage power lines
Generation Transmission (high-voltage) Hydro-Québec 2017 (André Gascon)
Design example
Context
10
Electrical Power Network : Control
Generation Transmission(high-voltage)
Distribution
PA PA PA. . . CED CED CED. . .
CCR
CCR : Centre de conduite du réseau
PA : Place d’affaire téléconduite (salle de téléconduite)
CED : Centre d’exploitation de distribution
1 provincial control center (CCR)7 regional control centers (PA)5 distribution control centers (CED)63 power plants (Generation)> 500 substations (Transmission)
Hydro-Québec 2017 (André Gascon)
Design example
Context
Tele
cont
rol
Ele
ctri
cal
Net
wor
k
HQ - ProductionHQ - TransÉnergie
HQ - TransÉnergie
HQ - Distribution
Loc
alR
egio
nal
Pro
vinc
ial
6
11
Electrical network control< local control room >
Local HMI
PA PA. . .
CCR
Hydro-Québec 2017 (André Gascon)
Design example
Context
< www.hydroquebec.com >
12
Electrical network control< remote control centers >
1 provincial center (CCR)7 regional control centers (PA)5 distribution control centers (CED)
Regional control center HMI (PA)
PA PA. . .
CCR
Provincial control center HMI(CCR)
Hydro-Québec 2017 (André Gascon)
Design example
Context
Journal de Montréal
TVA Nouvelles
7
SCADA : two channels
HMI
Automation systemProcess
uuu
Control, status
Alarms
Automatisms
and
Protection
Control
Active monitoring
Passive monitoring
Hydro-Québec 2017 (André Gascon)
Design example
13
Context
SCADA : Supervisory Control and Data Acquisition
Annunciator
Operator
14
Typical displays (SCADA HMI)
ADRSAD
ALB
Hydro-Québec 2017 (André Gascon)
Design example
Context
Planning and active monitoringControl and active monitoring
Passive monitoring
Annunciator
8
15
Annunciator panel
Hydro-Québec 2017 (André Gascon)
Design example
Context
Passive monitoring
<>
< www.Wikipedia.org >
Getting rare, but still in use
( Some automation subsystems can be in use for over 30 years )
16
Chronological annunciator (ALCID-SICC-I)
Getting rare, probably still a few in use
>>> Vraie capture, ou scan au besoin Hydro-Québec 2017 (André Gascon)
Design example
Context
( Some automation subsystems can be in use for over 30 years )
Passive monitoring
<>
9
17
Regional control center example (SAD)
In use from 1985 up to october 2005
Hydro-Québec 2017 (André Gascon)
Design example
Context
Passive monitoring
<>
18
Typical annunciator state machine
Hydro-Québec 2017 (André Gascon)
Design example
Context
Normal
Alarm
.
New unack
New ack
Return unack
Return ack
Acknowledge
Acknowledge
Reset
Field Alarm message
< ~ >
10
19
Context at Hydro-Québec
• Process control– Multiple interdependent variables - Dynamic– Risks - Conflicting goals– Typical task :
• Information Situation awareness – Anticipation Decision
• Hydroelectric power generation and transmission– Multiple processes; most of them fairly simple, a few quite complex– Complex configuration (and always changing); Complex HMS– Expert users; 24/7– Important risks; single contingency rule– Subject to numerous environmental influences : thunderstorms,
freezing rain, wind, temperature, solar activity, rain, river levels, ice, ice cover, forest fires, …
Hydro-Québec 2017 (André Gascon)
Design example
Context
20
Context at Hydro-Québec
• Expert users– Various backgrounds : electrical maintenance,
technician, outage coordinator, … (practically nobody with an engineering background)
• Then local operator, telecontrol operator, telecontrol dispatcher.
– Long training : master and apprentice, coaching– Lots of experience (often at least 10 years of network
operation for a provincial control dispatcher).
Hydro-Québec 2017 (André Gascon)
Design example
Context
11
21
Context at Hydro-Québec
• To get the right information for design
Hydro-Québec 2017 (André Gascon)
Design example
Context
22
Situation at Hydro-Québec
• Automation has steadily increased :– Number of operators has shrunk dramatically
– From 20 to 30 (up to 70) substations for a tele-control desk
• Number of alarms points rises sharply with automation :– More alarm points than before (from 30 to 300 alarm points for an
alternator, upward of 18,000 alarm points in the Beauharnois power plant)
– Cascades of up to 200 (sometimes up to 400-800) alarm messages at the beginning of a perturbation.
• Network operated much nearer of limits (maximize use of network capabilities)
Up to 3000 alarms a day for a tele-control desk.
< The number of phone calls during a shift reach 300 > Hydro-Québec 2017 (André Gascon)
Design example<>
12
23
Advance exercise : Questions Exercise
Hydro-Québec 2017 (André Gascon)
24
The necessary attitude
Be very careful with your own impressions,concentrate on user reactions,
try to understand their goals, know the context.
Don’t jump into design until you truly understand the task and the user’s goals.
“ The best attitude for today's designer to adopt is tothink of the user as a different species that one knows nothing about;
the user must be investigated scientificallyto determine the optimal design features to facilitate the use of interactive systems ”
Deborah Mayhew
Exercise
Hydro-Québec 2017 (André Gascon)
<>
Observe the user doing his job Analyze the task : understand the user’s goals Validatin : observe the user using the mock-up or prototype to
do his job
13
25
The necessary attitude
You are building a tool for somebodywho is doing a job you don’t really know about
(at least not at the skill level).
Skills
Rules
Knowledge
Users * Engineers
Exercise
Hydro-Québec 2017 (André Gascon)
On top of that, you know too much about computersto rely on your own impressions to design an HMI.
< analogy : driver vs mechanics in car racing >
Jens Rasmussen
* : experienced users
26
Observations (alarms)
• After a perturbation :– In a power plant, the operators go directly to the control panel, without first looking at the alarm messages on the screen– « Acknowledge – reset » without prior reading of alarm messages (the operator looses the historical information, gets the present state)
• High proportion (operators opinion) of one by one judgment about pertinence of alarms
• From 70% to 80% (operators opinion) of alarms are irrelevant« 30 alarms (then grouped on 12 indicator lights on the control panel) were enough for an alternator, why do we need 300 now ? »
• In important events, a high proportion of the alarms are momentary
• In some cases, it is difficult to understand the new state of the plant immediately after a complex event. The annunciator doesn’t seem to help.
• Not much pattern recognition for alarms in the regional control center *, only reading of alarm messages
* : except on the mosaïc overview display : changes of state (general state of the electrical network)
Hydro-Québec 2017 (André Gascon)
Design example
14
27
Techniques
Observation
Analysis
Usability tests
Design
At workplace, performing (or simulating) actual work (think aloud), one user at a timeNotes, screen captures, video camera (with sound), photos
Hierarchical Task Analysis (HTA), strategies, problems, errors
Optimize design with software experts
Use the mock-up to do real task examples (think aloud)(Users do their task while thinking aloud) (techniques as for observation)
Look everywhere for ideas
Iterate
HTA (goals, data) --> Data, grouping, sequence
Heuristic evaluation
Hydro-Québec 2017 (André Gascon)
Techniques
Situational awareness (anticipation)
28
Techniques : Hierarchical Task Analysis
Why ?How ?
HTA
First draft of HMI(interface and task)
Information needs
- What do you do ?- When, how often ?- How do you do it ?(ask the user to do the actions,or to simulate them)- What are you looking at ?- Why do you do it ?- Problems ? Errors ?- Strategies ?
GoalsActions
< data, grouping, sequence >
< chains of goals>
Hydro-Québec 2017 (André Gascon)
Techniques
bold + blue = main outputs
> Look at all the « tools »
During the interview, let the task go (don’t interrupt to obtain direct answers to your interrogations)
15
29
Techniques : Hierarchical Task Analysis
Users can validate the HTA
Hydro-Québec 2017 (André Gascon)
Techniques
In most cases, the tasks already exist in some form
New HMI revised task• main goals mostly intact• some sub-goals may change• some actions will probably change• function allocation (human vs machine) may change
30
High level understanding from task analysis
• During normal operation
– Situation awareness / Anticipation
– Periodic active monitoring
(period depends of the situation)
– Mostly passive monitoring of anomalies
Alarms managed one by one
Hydro-Québec 2017 (André Gascon)
Design example
SA
16
31
Situation awareness
• Perceive
• Comprehend
• Anticipate– Anticipation is testable
Hydro-Québec 2017 (André Gascon)
Design example
« Designing for situational awareness : an approach to user-centered design », M.R. Endsley, D.G. Jones, 2nd edition, CRC Press (Taylor & Francis), 2011
32
High level understanding from task analysis
• Perturbation– 1 : situation assessment (restoration of lost situation awareness)
• active gathering of state info; then action on the symptoms
• annunciator : at most a minor role (even if well designed) *
– 2 : preparation for return of service• annunciator : remaining alarms (« Secondary overview »)
• decide on a power restoration strategy
• historical log : may be useful to find causes; then action on the causes
* : a « dark-panel » low-key topographical annunciator is probably part of a better solution• « unauthorized change of state » indicators in a topographical diagram are useful (to highlightchanges in the state of the process)
Hydro-Québec 2017 (André Gascon)
Design example
Diagnosis is a different task, done after the fact• for important events, it is normally done by network engineers, not by operators• it requires a different tool, with filters, a search function, separate events for new alarm and return to normal, date and time (to the millisecond), …• often, the engineer will use an Excel spreadsheet to analyze the data
17
33
Literature and existing products
• Intelligent filtering (to reduce the number of alarms) is difficult to implement and extremely difficult to maintain
• Almost nothing on representation
• No tests with experts
• Situation awareness thought to be important
• Products : design seems neglected (ex.: date-time on the left, including year); looks as if design was done for diagnosis
Hydro-Québec 2017 (André Gascon)
Design example<>
34
Conjecture
• After an perturbation, the priority is to re-establish situation awareness; diagnosis is secondary.
• After a perturbation, one needs to update his mental model of the process before making decisions
“ The other important aspect of cognitive skillsin on-line decision making is that
decisions are made within the context of the operator's knowledgeof the current state of the process.
[and the anticipated evolution of the process].“Lisanne Bainbridge 1983
Hydro-Québec 2017 (André Gascon)
Design example
This remains true for most tasks done with a computer(state of the « process » and state of the computer program)
Decisions are based on anticipation
18
35
Annunciator Design
First priority after a perturbation:Understand the new state of the plant
HMI has to be designed for the perturbed situations, and must be in constant use * .
It must also be useful in normal operation
First goal after a perturbation :Safeguarding the network
Hydro-Québec 2017 (André Gascon)
Design example
We need to work on the representation of information
to help current state perception.
For complex systems, pattern recognition is probably
the best way (if not the only one).
* : must be already displayed when the event occurs, and
the operator must be very familiar with the display.
36
Techniques : Mock-up
First mock-up Paper and PostIt, or Visual Basic used as a drawing tool
Styrofoam and paper for physical aspects
Mock-up Prototype*
Evolution Usually Visual Basic used as a drawing tool, adding just enough functionnality to properly test specific sub-tasks
Never (almost) demo the mock-up to users **
Ask the user to use it to do actual work
The mock-up is the best communication tool between user, client, interaction designer and programmer
Designing the mock-up and doing usability tests enables one to push the interaction design much further
Other mock-up tools : PowerPoint, Excel, DHTML
(Excel is good for simple web site mock-ups)
* : a partial prototype may be needed in some cases ** : however, a demo may be useful for clients
Hydro-Québec 2017 (André Gascon)
Techniques
19
37
Mock-up example (Visual Basic) (SCADA HMI, SAD+A)
< amélioration possible : petit .avi avec bouton pour le démarrer > Hydro-Québec 2017 (André Gascon)
Techniques
Poste 1
Poste 1Poste 1
38
State annunciator (mock-up) (« Liste-État »)
Hydro-Québec 2017 (André Gascon)
Design example<>
20
Usability testing (« Liste-État »)
Caméra vidéo
commandes de l’annonciateur topographique
schéma unifilaire avec annonciateur topographique intégré (topo) (fonctionnel ou non selon la condition)
annonciateur à liste (liste-chrono ou liste-état)
ou
ou
puis
ou
3 scénarios 3 scénarios
Design example
39
<>
40
Visual Substation (commercial version)
Hydro-Québec 2017 (André Gascon)
Design example
< Cybectec, then Cooper >
<>
Poste 1
21
41
ALCID-II : new local annunciator
Hydro-Québec 2017 (André Gascon)
Design example<>
42
Regional control center example
Currently in use (2007 )
Color rectangles are flashing
Hydro-Québec 2017 (André Gascon)
Design example
Context
Design : mid-1990 (independent, some configuration by HQ)
<>
22
43
Provincial control center example
Currently in use (2000 ) Hydro-Québec 2017 (André Gascon)
Design example
Context
Design : mid-1990 (independant)
<>
44
Topographical annunciator (mock-up)
Hydro-Québec 2017 (André Gascon)
Design example
< substation level one-line diagram >
Tests : no advantage, distracting need state first(alarms are secondary; their representation should remain discreet)
Note on tests : the operators were not familiar with this type of display.
<>
23
45
Overview display (« Synoptique Réseau »)
Summer 2007
Hydro-Québec 2017 (André Gascon)
« I feel the network »
Design example<>
TVA Nouvelles
46
Overview display with alarms (dark-panel, low-key)
+ mesures à quartilesSpring 2009
Hydro-Québec 2017 (André Gascon)
Design example
< telecontrol center level one-line diagram >
< not done on the substation level one-line diagram (except for CENA) >
<>
anonymiser
24
Annunciators these days
Beauharnois : UCC + control panel
Telecontrol dispatcher desk : PTR + PTS
Provincial control dispatcher desk : several annunciators on monitors (transmission, automatisms, LIMSEL, state estimator) + mimic wall
>>> à compléter (+ dessins ou photos)
Hydro-Québec 2017 (André Gascon) 47
Design example
48
• Design example : Context Users Observations Conjecture Design
– Questions– Advance exercise : Context Results Recommandations
• Cognitive Systems Engineering
• Major lessons learned : Methodology and techniques Design
• Articulation with the rest of IT
25
Overall reasoningObjective : performance-efficiency of the enterprise in its mission
Means :- Performance of individuals in their tasks- Collective performance : business processes, communication
Tools :- IT tool HMI : information, representation/layout, display HW- Direct human interactions- IT services and functions
IT requirements difficult to define for complex systems, especially when users are experts
- human beings are part of the system- tasks- processes
Hydro-Québec 2017 (André Gascon) 49
N.B. : Standards, guidelines and good practices are very useful, but are far from being sufficient to garantee a good design.
Cognitive Systems
Engineering
50
Specific expertise
• Cognitive Systems Engineering methodology
+ Integration with requirements engineering and software engineering methodologies.
• Techniques to find the necessary information– Interviews, task analysis, usability testing, …– < how to get access to the user (intuitive) expertise {in their task} >
• Knowledge of human being : (applied psychology)– Vision, perception, situation awareness, decision making, mental load,
nature of expertise, human error and biases, representation effect, Gestalt, …
• Display technology, HMI design, interaction design, and business processes description
Hydro-Québec 2017 (André Gascon)
Cognitive Systems
Engineering
Design methodology and techniques, validation techniques
26
Demands, needs, design and validation
« In any case, what users want and what users need are two different things, which is why it’s long been a primary usability guideline to watch what users do, rather than listen to what they say. »
« Over the past 25 years, work in usability has shown that one of the best ways to evaluate a design’s quality is by watching users interactwith it (through either a functional or mocked-up screen). Again, if years go by before the developers do this, most of their development effort will have been spent producing the wrong design. »
Jakob Nielsen < www.useit.com >
Hydro-Québec 2017 (André Gascon) 51
Cognitive Systems
Engineering
Demands, needs and requirements
( client/user vocabulary )
( client/user vocabulary )
( IT vocabulary )
Demands ≠ Needs (*)
Requirements = Translation of needs
Demands
Needs(IT)
Requirements (IT)
Demands formulated by the client, often expressed as solutions
Description of the design of the business solution, as much as possible validated
Items of the contract for building/integration/configuration by IT
This is a major succes factor. Preferably done via usability testing.
* : Give satisfaction to the client ≠ Satisfy all his demands
What the client asked for
What the client really needs (most of the work of the business analyst is to define those needs).
The order given to IT
Concepts
52
< High-level business needs >
27
53
Cognitive Systems Engineering : fundamental concepts
Definition of (IT) needs (≠ gathering of needs)Through the design and the validation of the HMI.
DesignBased on task analysis and usability testing.< ie.: based on the users’ expertise in their task >
With technical optimization.
The design is expressed as a mock-up which is also the basic tool for usability testing.
Interaction design Functional requirements specification
Hydro-Québec 2017 (André Gascon)
Cognitive Systems
Engineering
Analogy : « I want to hang this painting on this wall », . . .< what do you need to achieve your task goals ? >
Nota bene : Validation vs Verification
Validation
« Confirmation that the product or service, as provided (or as it will be provided), will fulfill its intended use.
In other words, validation ensures that “you built the right thing”. (See also “verification”.) »
Verification
« Confirmation that work products properly reflect the requirements specified for them.
In other words, verification ensures that “you built it right”. (See also “validation”.) »
CMMI® for Development, Version 1.3, SEI
Hydro-Québec 2017 (André Gascon) 54
Cognitive Systems
Engineering
28
55
Methodology : main actors
Understand the task
Design - optimization
Validate the design
User expertise
Describe their actions
Do their task using the new tool (mock-up)
Ergonomist + software experts
Ergonomist
Ergonomist
Ergonomist
Ergonomist
Users
Hydro-Québec 2017 (André Gascon)
Cognitive Systems
Engineering
56
Methodology : design cycle
Gathering of information on the task
Design with optimization
Usability tests
Definition of needs = Validated design
Hierarchical Task Analysis (HTA)Information needs / sequence
Feasability analysis
Mock-up Functional requirements spec.
Activity Document
(with users)
(with users)
(with computer specialists)
Iterations
Iterations
Iterations
HMI : interactionServices / functionsDB
Hydro-Québec 2017 (André Gascon)
Cognitive Systems
Engineering
29
57
Methodology : « Double Diamond »
Finding the rightproblem
Finding the rightsolution
Explore the fundamental issues
Converge upon the real underlying problem.
Explore a wide variety of solutions
Converge upon a good solution
« Good designers never start by trying to solve the problem given to them : they start by trying to understand what the real issues are. »
« The Design of Everyday Things », Don Norman.
Design Process
58
Methodology and design process
• Design process
Hydro-Québec 2017 (André Gascon)
Design process
30
59
• Design example : Context Users Observations Conjecture Design
– Questions– Advance exercise : Context Results Recommandations
• Cognitive Systems Engineering
• Major lessons learned : Methodology and techniques Design
• Articulation with the rest of IT
• « End »
• Additionnal information
60
Major lessons learned : methodology
• Demands ≠ Needs ( needs need to be « defined » and validated )
• Task mode : to extract information useful for design– < task observation, usability testing >
• Mock-up : as a design tool, as a medium for usability testing, and as a communication tool– < + essential for dynamic aspects, including visual momentum >
• Main objective : performance of the human-machine system ( the performance of the IT system is not an objective per se )
• Usability testing and iterations are absolutely necessary
• Design for situation awareness : a very powerful concept
Methodology and techniques
31
61
Design process
The study of the task (analysis and synthesis) guides the design
The mock-up supports the creative process, provides a representation necessary to make progress (in the design)
Usability tests enable the designer to validate his design, to uncover design errors, to continue to make progress
Iterations are necessary because we never get it right the first time
Observation & analysisStrategies Errors Problems
Task analysis UsersUsability tests
Design of HMI and new task
Objectives are set for the « human-machine system » (performance centered)
Hydro-Québec 2017 (André Gascon)
Methodology and
techniques
<>
( Interaction design )
62
“ Most (75%) of the ergonomist’s work is to describe
and understand what already exists. ”“ ... then everything will fall into place … ”
Jean-Marc Robert
+ any idea, however brilliant it may look, has to be tested.
Good usability tests are essential.
Methodology
Major lessons learned : Methodology
Good functionnal (not aesthetics) design is often not noticed by users.They simply won’t complain about the design.
32
63
Situation awareness
• Perceive
• Comprehend
• Anticipate– Anticipation is testable
Hydro-Québec 2017 (André Gascon)
Situation awareness
« Designing for situational awareness : an approach to user-centered design », M.R. Endsley, D.G. Jones, 2nd edition, CRC Press (Taylor & Francis), 2011
64
The fundamental objective is correct anticipation
During tasks, and during idle time Situation awareness
Correct anticipation
Situation awareness analysis :
- What are you anticipating ? At what time scale ?
- Why ? (information, reasoning, expertise, . . .)
- Test : What do you anticipate ?
Hydro-Québec 2017 (André Gascon)
Situation awareness
33
65
Design for situation awareness
Information available in parallel
- Peripheral perception
- Minimal size - simultaneous localization and mapping
- Gestalt
> D.S. Tan, E.R.Tufte, M. Wertheimer : see references
> Process control, diagrams, schematics
> Can navigation within an application be considered a spatial task ?
Hydro-Québec 2017 (André Gascon)
66
• Design for situational awareness
• Design layout for peripheral perception
• Large screens are beneficial for spatial tasks
Design strategies
Hydro-Québec 2017 (André Gascon)
Some lessons learned : Design
34
67
• Design for situational awarenesso State of the process / data / document . . .
o State/mode of the application / computer
So the user can anticipate correctlyAnticipation is testable
• Aware :– Structure / « logic » of mental model coherent with application /
computer
– Enough information to ensure that mental model is up to date
– Applies broadly : aware anticipate decision
Design strategies
Hydro-Québec 2017 (André Gascon)
Some lessons learned : Design [1]<>
68
• Design so that the user can anticipate correctlywhat the software will do :– Coherence, within the application, and with interaction
concepts in general use.
– Affordance
Design strategies
Hydro-Québec 2017 (André Gascon)
Some lessons learned : Design [1a]<>
• Questions you can ask– What are you anticipating ?
– What are you monitoring ?
– What are your clues ?
35
69
• Experiment with a sheet of paper< Central perception >
< Peripheral perception >
• Reading experiment< \pro-ics\Emploi-Planif\Evol-SCADA\ExperienceLecturePerceptionPeripherique-1.pptx >
Design strategies
Hydro-Québec 2017 (André Gascon)
Peripheral perception<>
70
Peripheral perception is important for pattern recognition and navigation
– Design layout for peripheral perception : micro / macro design
• Know Gestalt and the characteristics of perception via peripheral vision . . .
• Density not important per se (at least for expert users)
• « Good » layout is critical
• Beware of blinking or moving objects
Design strategies
Hydro-Québec 2017 (André Gascon)
Navigation is a spatial task . . . Visual momentum involves peripheral vision . . .
Some lessons learned : Design [2]<>
Peripheral perception pattern recognition situational awareness
Some caracteristics of peripheral perception :- “Black & white” - Detection of alignments, regularities, Gestalt- Very sensitive to movement
36
71
Absolute size of screen is a factor for spatial tasks
Design strategies
Hydro-Québec 2017 (André Gascon)
Some lessons learned : Design [3]<>
Exocentric
(outside of my space)Egocentric
(within my space)
[ SLAM (body in its environment ]
(very robust, very well trained)
SLAM : Simultaneous Localization and Mapping
For electrical network diagrams, minimum size 90’’ diagonal, minimum distance 80’’
72
• Advance exercise : Context Results Recommandations
• Design example : Context Users Observations Conjecture Design
• Cognitive Systems Engineering
• Major lessons learned : Methodology and techniques Design
• Articulation with the rest of IT
37
73
Business solution = business process (and tasks)
Design strategies
Hydro-Québec 2017 (André Gascon)
Business Analysis and Cognitive Systems Engineering
Gather / Inquire> Objectives
Bottom-up analysis : from the individual tasks
Gather / Inquire> High level requirements> Business rules
Interviews / observation> Task analysis> Strategies/Errors/Problems> Information
Inquire> Current data
Model> Current business process
Identify problemsFind causes
Future business solution
Model the future business process
Model the future tasks
Design and validate HMI
Describe UC
Define future data
Requirementsspecification
Business solution
IT solution
EPC : Event (driven) Process Chain
Event
Action
Actor
Applic.
Inform.
74
« Un pont entre les Affaires et les TI » (CS et AA)
Analystes d'affaires
RelationsActivités
Spécification des exigencesCueillette des demandes
Design d'interaction / définition besoins
Traduction des besoins (clients) en exigences (TI)
Discussion de la solution
Stratégie d'essais fonctionnels
Évaluation des demandes de maintenance
Affaires
Gestion clientFiches projet
Suivi TI (projets et DT)
Plan d’affaires
Processus d’affaires
Business Case
Pilotes
Problématiques
Intrants aux fiches projet
Demandes / besoins
Intrants au Business Case
Essais
UtilisateursAnalyse de tâche
Essais d’utilisabilité
Processus d'affaires des clients (documentation, objectif à long terme)
Portefeuille de projets TI
Portefeuille de maintenance
Avis de conformité
Bureau d’architecture
Priorisation projets [via fiches projet]
Architecture-cible
Bilan de santé
Projets
Comité de coordination
Accompagnement
Architecture projet
Plan de projet / Recommandation
Avis de conformité
Analyse préliminaire
Réalisation - relation réalisateur
Essais d’acceptation
Business Case (longitudinal)
TI
Maintenance
Maintenance évolutive, corrective, pérennité et adaptative demandant code.
Carnet de maintenance
Demandes de transformation
Priorisation DT / calendrier liv.
Objectifs d’entreprise
Orientations stratégiques
Vigie affaires
Suivi TI
Suivi bénéfices (BC longitudinal)
Conseiller stratégique
Gestion entr.
Orientation et Planification
Visibilité
Évaluation d'opportunité
Analyse préliminaire
Analyse d’affaires
Vert : CS impliqué dans l’activité
Bleu : AA impliqué dans l’activitéExploitation Opérationalisation
Dépannage
Transversal : inter-projet, inter-domaine
Longitudinal : évolution, vigie, …
OP
SÉv
olu
tion
Expl
oita
tion
Hydro-Québec 2017 (André Gascon)
<>
38
75
Four essential ingredients for a successful project- Business process analysis, strategic alignment (, ÿ )- Interaction design (display devices, appearance et behaviour)- Technical design- Project management
Definition of needs (functionnal)
- Demands ≠ needs
- Needs = Interaction design
- It is essential to validate the design
- Tasks and processes
Elements of a successful IT project
< Interactive systems >
< Objective = profitability (max perf/cost ratio) >
Hydro-Québec 2017 (André Gascon)
Articulation with « the rest of IT »
Task observation and analysis is the foundation
for both interaction design and business processes analysis
76
CSE = more than the Human-Machine Interface (HMI)
Role of Cognitive Systems Engineering
Cognitive Systems Engineering Software Engineering
Requirements specification
Task (and business processes) design Functions, services, data
SE : insure technical quality, control costs and delays
CSE : insure tool will improve the performance of the HMS
> Help the clients define their needs <
Hydro-Québec 2017 (André Gascon)
Articulation with « the rest of IT »
<>
39
Requirements specification
Description(section 2)
Contexte
Design opérationnel / interaction
Principales contraintes
AA
AA
AA Gouvernance, ATI, Exploitation
Exigences fonctionnelles AA : énoncées
Contraintes
ATI : énoncéesExig. archi. fonctionnelle
ATI : énoncéesExig. archi. technologique
ATI : énoncéesExig. sécurité
Expl. : énoncéesExig. exploitation
Gouv. : énoncéesExig. gouvernance
AA : annoncées(fait référence à un document distinct)
Exigences(section 3)
Contexte : organisation, processus, environnement, objectifs, …Opérationnel : 90% pour clientContraintes : très peu pour le client
Besoins
Exigences
Utilisation T110
Tous les sujets, et « n/a » si non pertinentEnvergure selon le contexteNuméros P+ dans les titres de section
77
Description de l’interaction : tâches et IHM (dispositifs, apparence et comportement)
Hydro-Québec 2017 (André Gascon)
Table of contentsAdditional Info
<>
78
BA - UX
Functions (services) / data
Interaction
Task analysis
Business process analysis
Corporate objectives
Corporate strategies
Corporate structure
BA
UX
Hydro-Québec 2017 (André Gascon)
Design levels
Interaction design
Functional design
Operational design
Requirements
Requirements
Constraints / Ideas
Constraints / Ideas
Business AnalysisUser Experience
Articulation with « the rest of IT »
Business needs (high-level)
<>
40
Niveaux de design
** : allocation des fonctions : humains, système TI, autres systèmes TI
Représentations / modèles
(typiques) (en plus du texte)
EdPC / BPMN
OCD +
SCD
stm sd act bdd
. . .
Tâches
IHM
SFD*
AHT
Maquette
Descriptions
ucdSFD*
SFD*
Protocole
Design
Organisation
Processus
Opérationnel
Allocation **
* : SFD : services, fonctions, données
Interaction
Fonctionnelle
Arc
hite
ctu
reComposants
Besoins p/r au système TI
Exigences fonctionnelles sur le système TI
Contraintes sur le design du système TI(Arch., Séc., Expl.)
Besoins opérationnels
Sect. 2
AA
AT
Besoins et exigences
H140
Sect. 3
Sect. 3
. . .
ad hoc
En
viro
nn
em
ent
Sys
tèm
e T
I
AHT
IHM
UC UC UC
ucd : use case diagramact : activity diagramsd : sequence diagramstm : state machine diagrambdd : block definition diagram
EdPC : event driven process chainOCD : operational context diagramSCD : system context diagramAHT : analyse hiérarchique de tâche
IHM : interface humains-machine : apparence et comportement, ainsi que dispositifs
*** Sect. 2-3
*** : si exigences sur les traitements ou données
Technologique
Sécurité
H140 : spécification des exigences(propriétaire, utilisateur, réalisateur)- Section 2 : description- Section 3 : exigences
CS : Conseiller stratégiqueAA : Analyste d’affairesAT : Architectes (fonctionnel, technologique et sécurité)
+ : autres au besoin, comme stm pour automatismes par exemple 79
Orientations stratégiques
CS
Hydro-Québec 2017 (André Gascon)
Articulation with « the rest of IT »
<>
80
Éléments fondamentaux de LEAN LEAN
Obsession continuelle
« Kaizen »Amélioration
continue
Minimiser le gaspillage (ce qui ne produit pas de valeur pour le client)Maximiser la valeur pour le client
Régularité du flux : stabilité de la capacité et de l’expertise
Cette valeur est elle-même subordonnée à la valeur de l’activité (supportée par l’outil TI) pour les clients de TransÉnergie et d’Hydro-Québec (entreprise)
Flux tirés (« Pull ») : chaîne de « clients » internes aux TI, jusqu’au client de TI lui-même; puis jusqu’aux clients de TransÉnergie et d’Hydro-Québec.
Identifier la valeur pour le client Connaître les processus et tâches TI; les ajuster au besoin
< activité en continu > < activité en continu >
Fondations : terrain (« Gemba »), et analyse de tâche
1 C0
Eff
icac
e
Eff
icie
nt
<>
41
CSE vs frameworks and books of knowledge
81
“Frameworks” and “Books of Knowledge” are usually meant to accommodate any design methodology.
One must first choose a design methodology to make relevant choices in a framework or a book of knowledge.
For software with a human interaction aspect, the best design methodology is probably the “user centered design” approach of the ISO-9241-210 standard (ex-13407).
[ CMMI, Macroscope, Agile, BABoK, PMBok, SwEBoK, ... ]
Hydro-Québec 2017 (André Gascon)
Articulation with « the rest of TI »
82
Some references
Dreyfus, H., Dreyfus, S., (1986), « Five steps from novice to expert », pp. 16-51 (chap. 1) in Mind over machine, The Free Press, New York.
• Gascon, A. et, Robert, J.-M., (2003), « L’ingénierie cognitive au service de la conception d’interfaces humains-machines : leçons tirées de 8 années de pratique en entreprise », Actes du 5e Congrès international de génie industriel, 2003.
Tullis, T.S., (1993), « Is user interface design just common sense ? », Proceedings of HCI International '93 Conference, Orlando, FL, Aug. 1993, V 2, p. 9-14, Elsevier Science Publishers.
• Wickens, C.D., (1992), « Engineering Psychology and Human Performance », Harper Collins.
• Endsley, M.R., Jones, D.G., (2011), « Designing for situational awareness : an approach to user-centered design », 2nd edition, CRC Press (Taylor & Francis), 2011
• « CITATION.doc »
Hydro-Québec 2017 (André Gascon)
42
83
Some references
• Kuai, S.-G., Yu, C., (2006), « Constant contour integration in peripheral vision for stimuli with good Gestalt properties », Journal of Vision (2006), 6, 1412-1420.
• Rosenholtz, R., Li, Y., Nakano, L., (2007), “Measuring visual clutter”, Journal of Vision (2007), 7(2):17, 1-22.
• Tan, D. S., Gergle, D., Scupelli, P., Pausch, R., (2006), “Physically Large Displays Improve Performance on Spatial Tasks”, ACM Transactions on Computer-Human Interaction, Vol. 13, No. 1, March 2006, Pages 71-99.
• Tufte, E. R., (1995), “Envisioning Information”, Graphics Press. : chapter 2 : Micro / Macro Readings, especially pages 37, 38, 50 and 51.
• Wertheimer, M., (1923) , “Laws of Organization in Perceptual Forms”, translation published in Ellis, W. (1938) A source book of Gestalt psychology (pp. 71-88), London : Routledge & Kegan Paul. [ old, but charming , and quite useful for me ]
Hydro-Québec 2017 (André Gascon)
84
EndQuestions ?
Additional information• Mock-up examples• Cognitive Systems Engineering definition• More on lessons learned• Difficulties• Suggestions• My opinion on the future of Cognitive Systems
Engineering• More on process control• More context information for advance exercice• How I got to CSE• Requirements specification table of contents• Industrial engineering and CSE
Hydro-Québec 2014
43
85
Mock-up examples
…\ICS\Pres-ICS-IdT-2005\Exemples-Realisation-ICS.ppt
Hydro-Québec 2017 (André Gascon)
Techniques
3, 5 : Beauharnois
8 : ALCID-II
12, 13 : SEQAM
14 : ADR
17 : étiquetage
19 : revue Transport CCR
20 : procédure aidée par ordinateur
21, 22 : délestage cyclique
28 : DREX
35 : pupitres CCR
39, 40 , 41, 42 : synoptique réseau
44, 45, 46 : PG&E
86
Cognitive Systems Engineering
• « Discipline that : – Takes an interest in all stages of the life of complex
human-machines systems (« Cognitive Systems »)– Calls for knowledge and methods of many nature,
social and human science disciplines– Has the fundamental objective to improve the
performance of human-machine systems »
Jean-Marc Robert
......
Hydro-Québec 2017 (André Gascon)
44
87
Lessons learned• Analyze work activities (and use HTA to design the first draft of the HMI)• Don’t consider demands as needs; try instead to uncover the real needs
– Beware of needs expressed as solutions• Insure that the user is in « task mode » to obtain the right information• Don’t ask users to validate the requirements specification• Beware of the pitfalls of participative design• Keep contact with the field by regular visits
• Evaluate the HMI according to the task logic and the performance of the human-machine system• Do not aim for user-friendliness as a primary objective
• Build mock-ups
• Do the complete Cognitive Systems Engineering design cycle
• Explain the design process to the decision makers to convince them of the merits of the recommended design.
• One way to help convince project leaders, expert consultants, client representatives, …, is to tell stories about users performing their tasks. Seeing a video of users in their activities may also be useful.
– describe the main problems encountered by the users and their consequences ($)– describe observations and the actual design process
Task Mode
Mock-up
Objectives
......
Hydro-Québec 2017 (André Gascon)
88
Difficulties
• Solutions becoming needs, or needs expressed in the form of solutions– Technical « trips » : solutions in search of a problem
• CSE results often difficult to explain to software developpers (in fact to anybody who didn’t spend much time observing users at work); also often difficult with ex-users : they don’t feel the need to check with users, they stick with their own impressions.
• Many computer specialists rely on their own impressions, and think they can design good HMI.
• Most managers don’t see the importance of a good HMI design (enabling performance of users in their tasks, not only « user-friendly »), or of a bigger monitor, for that matter.
• Difficult to convince project leaders to accept the time and cost of CSE• Usability testing limits with experts
......
Hydro-Québec 2017 (André Gascon)
45
89
Suggestions• When a client demands something, find out why, find out what are
the task objectives
• Pay regular visits to users; be also there when the software is installed Stay « grounded » in the reality of the field
• Always check with users (3-4 at least, if possible); but don’t let them do the whole design; don’t rely on your own opinion
• Never (almost) do demos to users, instead ask them to use the product to do actual work
• Be rigorous (do a complete CSE cycle), not necessarily formal
• < + citations.doc >
......
Hydro-Québec 2017 (André Gascon)
90
CSE future (my opinion)
Still far from complete generic solutions bottom-up analysis is essential
Compatibility with human being
Compatibility with task
Importance of details Nature of expertise
Articulation with business processes and the structure of the organization
Better representations : better adapted to human beingsBetter interaction means : better adapted to human beingsUser aware of what the computer is doing {Human-Computer Cooperation (Norman)}
Business Analysis
......
Hydro-Québec 2017 (André Gascon)
46
91
HMI design for process control : future (my opinion)
Process ControlTasks
95% = acquisition of information, judgment on the quality of the information; importance of presentation
5% = decision
Monitoring better situation awareness, without saturation (especially the rapid restoration of situation awareness following a complex event)
( up to date mental model of the « process + computer » )
? Design for peripheral perception ?
......
Hydro-Québec 2017 (André Gascon)
92
Advance exercise : some context information
Dimly lit room, dark panelOther windows = color on black
Capture of momentary alarmsAlarm management(« son », « acq », « rap »)
Alarms : part that is not automated
(2 steps not to miss alarms)
- Undo not available
- Locate
- Mock-up is not perfectSeparate « New » and « Return to normal »
messages useful in log (for analysis after the fact)< but not in real time >
Exercise
Hydro-Québec 2017 (André Gascon)
47
93
How I got to Cognitive Systems Engineering
• Physicist
• Programmer-analyst (real-time software)• Project leader (automation software)
• Requirements specification, automation of the Beauharnois hydroelectric power plantCognitive Systems Engineering to define (IT)
needs
......
Hydro-Québec 2017 (André Gascon)
94
How I got to Cognitive Systems Engineering
One-line diagram of the a power plant
......
Hydro-Québec 2017 (André Gascon)
SICC-I console (no longer in use)
48
Sections du T110 et documents P+
1. Introduction
1.1 Portée du document
1.2 Audience visée
1.3 Portée de l’outil < P140 >
1.3.1 Résumé des fonctions < P250 >
1.3.2 Applications de l’outil < P140 ? >
1.4 Définitions, acronymes et abréviations
1.4.1 Termes et acronymes spécifiques à Hydro-Québec
1.4.2 Termes et acronymes d’usage courant
1.5 Références
1.6 Vue d’ensemble du document
95
Sections du T110 et documents P+
2. Vue d’ensemble de la solution d’affaires2.1 Contexte du système / projet < P140 >2.1.1 Raison d’être du système < P140 >2.1.2 Situation actuelle < P120 >2.1.3 Demandes < P100 >2.1.4 Sources d’exigences2.1.5 Présomptions et dépendances2.2 Description générale de la solution d’affaires < P140 > < P240 >2.2.1 Objectifs du système humains-machines < P140 >2.2.2 Relations à l’entreprise et aux processus d’affaires < P140 >2.2.3 Interaction humain-machine < P240 >2.2.4 Interfaces < P240 >2.2.5 Fonctions majeures < P250 >2.2.6 Structure des données < P170 >2.2.7 Contraintes < P261 >
96
49
Sections du T110 et documents P+
3. Exigences spécifiques3.1 Interfaces externes3.1.1 Interfaces système < P >3.1.2 Interfaces utilisateurs < P240 >3.1.3 Interfaces matérielles < P261 >3.1.4 Interfaces logicielles < P261 >3.1.5 Interfaces de communication < P261 >3.2 Fonctions < P240 > < P250 >3.2.1 Fonction …3.3 Performance < P240 >3.3.1 Capacité (statique)3.3.2 Performance dynamique3.4 Base de données < P240 >3.5 Contraintes < P261 >
97
Sections du T110 et documents P+
3.6 Attributs du système TI3.6.1 Fiabilité < P240 > < P261 >3.6.2 Disponibilité < P240 > < P261 >3.6.3 Sécurité < P240 > < P261 >3.6.4 Maintenabilité < P240 > < P261 >3.6.5 Adaptabilité < P261 >3.6.6 Pérennité < P261 >3.6.7 Portabilité < P261 >3.6.8 Compatibilité < P261 >3.6.9 Autre attribut …3.7 Autres exigences3.7.1 Structure de l’information < P170 >3.7.2 Architecture logicielle < P261 >3.7.3 Structure des traitements et interfaces < P250 >3.7.4 Stratégie d’implantation < P261 >3.7.5 Stratégie de changement organisationnel < P140 >3.7.6 Impacts < P140 >3.7.7 Coûts et bénéfices < P140 >3.7.8 Infrastructure technologique < P261 >3.7.9 Langue < P240 >3.7.10 Support < P261 >3.7.11 Autre exigence particulière … (efficience, …)3.8 Groupes d’essais < P410 >
98
50
Sections du T110 et documents P+
4. Annexe : Registre des raisons de conception5. Annexe : Essais fonctionnels6. Annexe : Essais de performance et de limites7. Index
99
100
Cognitive Systems Engineering
• Industrial engineering = performance and efficiency
– Better tools better performance in task (speed, errors, quality, …)
– Applied to software (considered as a tool) = performance and efficiency of the Human-machine system (HMS)
– Applied psychology : because a human being is part of the system
Cognitive Systems
Engineering
Hydro-Québec 2017 (André Gascon)
<>
51